Introduction

Although corporate governance and corporate social responsibility (CSR) reporting have separately established themselves as well-researched areas, relatively less attention has been paid in setting up a link between these two. Since CSR disclosure is influenced by the choices, motives and values of those who are involved in formulating and taking decisions in the organisations, consideration of corporate governance mechanisms, in particular, ownership structure and board composition could be important determinants (Gibbins et al. 1990; Haniffa and Cooke 2005). The scant research in this area (for example, Johnshon and Greening 1999; Jo and Harjoto 2011) have predominantly found CSR choice and performance to be positively associated with internal and external corporate governance mechanisms, such as board independence, board leadership and institutional ownership. Also, available empirical evidence suggests that both corporate governance mechanisms and CSR are positively related to the market value of the firm (Beltratti 2005). Institutional theory proposes that corporate governance mechanisms are sometimes adopted for the purpose of gaining legitimacy (Biggart 1991; Hamilton and Biggart 1988). Also, CSR literature suggests that the need to allay concern over threats to organisational legitimacy have largely acted as a potent driving force for such disclosures (for example, Chen et al. 2008; Deegan et al. 2002; Rahaman et al. 2004). Therefore, a strong relationship between CSR and corporate governance mechanisms can be envisaged.

One important gap in both corporate governance and CSR literature is the paucity of such research in the context of emerging economies. Judge et al. (2008), in a cross-country research, identify the institution of law and order and less prevalent corruption as antecedents to corporate governance legitimacy. This is consistent with Uddin and Choudhury (2008), who point out that the traditional societies in developing economies, characterised by family dominance, corruption and political interference, are not conducive for the adoption of western-styled rational corporate governance models. However, despite this, due to institutional pressures mainly exerted by external aid agencies, many emerging economies across the world have adopted the Anglo-Saxon model of corporate governance, and the effectiveness of such corporate governance mechanisms in the context of such countries have been questioned [for example, India (Mukherjee-Reed 2002); South Africa (West 2006); South Korea (Reed 2002); Bangladesh (Siddiqui 2010)]. Therefore, the influence of corporate governance mechanisms on CSR disclosures may be different in emerging economies. This provides the context for the study.

For the purpose of this study, we investigate the effect of corporate governance mechanisms on CSR in Bangladesh during the period 2005–2009. Like many other emerging economies,Footnote 1 Bangladesh has also adopted a western-styled corporate governance model requiring greater board independence, separation of the CEO and the chairman, provision of audit committees, etc. In more recent times the Securities and Exchange Commission of Bangladesh (SEC) has issued a ‘corporate governance notification’ (SEC 2006) consisting of some guidelines in regards to corporate governance practices by the listed companies. A number of prior studies (for example, Sobhan and Werner 2003) report that due to family dominance, corporate governance mechanisms such as presence of independent directors in the board tend to be largely ceremonial. Moreover, recent studies on CSR in Bangladesh have identified family ownership as one of the major factors for the dismal level of CSR disclosures (Belal and Owen 2007). Given the traditional nature of its society (Uddin and Choudhury 2008), the efficacy of such corporate governance mechanisms, therefore, can be doubted. Bangladesh thus presents an interesting case to examine the relationship between corporate governance and the extent of CSR disclosures.

Our results indicate that in the context of Bangladesh, although managerial ownership generally has a negative relationship with the level of CSR disclosures, such relationship is significant and positive for export-oriented industries, suggesting that managers in export-oriented firms may have external pressures to provide more CSR disclosures. We also find greater public and foreign ownerships to have positive impacts on CSR disclosures. In the context of Bangladesh, we find internal corporate governance mechanisms such as board independence, and presence of audit committees to have positive influences on CSR disclosures. However, we do not find any significant impact of CEO/chair role duality on the extent of CSR disclosures. Overall, our results suggest that despite ownership concentration, corporate governance mechanisms involving independent outsiders are effective in influencing CSR practices in emerging economies. We also find firm age, profitability and size to be important determinants of levels of CSR disclosures in Bangladesh.

Our study makes a number of important contributions to the existing literature. By providing evidence that managers in export-oriented industries, despite high ownership concentration, have enough incentives to make high levels of CSR disclosures, we offer empirical support to prior studies such as Belal and Owen (2007) and Islam and Deegan (2008), who mention pressures from international buyers as a crucial factor for CSR disclosures in Bangladesh. Our results also imply that companies with high foreign ownership (and with foreign nationals in the board) report more CSR disclosures as a proactive legitimacy strategy to satisfy ethical foreign investors and to get more foreign capital (Haniffa and Cooke 2005). Overall, this suggests that like many developed economies, managers in emerging economies such as Bangladesh also use CSR disclosures as a strategic tool. Despite doubts over the role of independent directors in Bangladesh, we find a positive relationship between board independence and the level of CSR disclosures. Also, presence of audit committee is found to have significant positive relationship with CSR disclosures. However, the effectiveness of separation of CEO and chair as a corporate governance mechanism appear to be affected by the culture of selecting these two positions by members of the same family. Overall, our results seem to suggest that the presence of sound corporate governance mechanisms incorporating more outsiders in the board and different committees, albeit rare in the context of Bangladesh, seem to alleviate some concerns regarding the influence of family dominance on CSR disclosures. This may have important policy implications, as our results indicate that corporate governance mechanisms such as CEO chair duality are likely to be ineffective in the context of family-firm dominated economies such as Bangladesh, if at least one of these individuals is not required to be independent. We also extend the CSR literature by providing empirical evidence of determinants of CSR disclosures in an emerging economy context.

The rest of the paper is structured as follows. The next section elaborates the institutional context and existing CSR practices in Bangladesh. “Theoretical Framework” elaborates the legitimacy theory based justifications for corporate governance and CSR. This is followed by a section which reviews related literature and develops hypotheses. Subsequent sections then discuss the research design and empirical results. The final section concludes the paper.

Institutional Background and CSR Practices in Bangladesh

Like many other developing countries [for example, Jackling and Johl (2009) report that 60 % of the top companies in neighbouring are run by families; similar family dominance is reported other developing economies such as Turkey (Ararat and Ugur 2003), South Korea (Joh 2003), Taiwan (Yeh et al. 2001), Thailand (Wiwattanakantang 2002)], most companies in Bangladesh are either family-owned or controlled by substantial shareholders (corporate group or government). Farooque et al. (2007) report that, on average, the top five stockholders hold more than 50 % of a firm’s outstanding stocks. The paper states that managements in many companies are effectively just extensions of the dominant owners—they are closely held small- and medium-sized firms where corporate boards are owner driven. Consequently, most of the companies have executive directors, CEO and chairman from the controlling family. A survey conducted by Sobhan and Werner (2003) found that an overwhelming majority (73 %) of the boards of non-bank listed companies were heavily dominated by sponsor shareholders. This is consistent with other studies such as Reaz and Arun (2006) and Uddin and Choudhury (2008) that report similar family dominance in the board of directors. Imam and Malik (2007) analyse the ownership patterns of 219 companies from 12 different industries listed on the Dhaka Stock Exchange, the premium stock exchange in the country. It is reported that, on average, 32.33 % of the shares are held by the top three shareholders, the results being even higher for real estate, fuel and power, engineering, textile and pharmaceutical sectors. Uddin and Choudhury (2008) report that unlike developed economies, the influence of institutional investors is almost absent in Bangladesh as such investors mostly operate as secondary market traders. Siddiqui (2010) identifies institutional weaknesses, such as highly concentrated ownership structure, lack of shareholder activism, presence of a weak capital market, absence of second-order institutions and poor legal structure as potential impediments towards the adoption of western-styled corporate governance models. This is consistent with Uddin and Choudhury (2008) who mention that the traditional nature of the Bangladeshi society, characterised by family dominance and informal control mechanisms, may not be suitable for rational corporate governance models. This is consistent with Chen and Nowland (2010), who find the corporate governance practices followed by family-owned enterprises in Asia to be ‘well below the optimal levels’ (p. 4).

As is the case in many other developing countries, a key strategy for achieving faster economic development in Bangladesh is one of rapid industrialisation. The industrial sector now makes a significant contribution to the economy of Bangladesh.Footnote 2 With that objective in mind, a private sector-led industrial development policy is being aggressively pursued with the aim of attracting as much foreign investment as possible. While this strategy clearly makes some economic sense, it has at the same time created many significant adverse social, ethical and environmental effects. This in turn has led to increasing demands for enhanced accountability and transparency in business practices (Belal and Owen 2007).

The Companies Act of Bangladesh (GoB 1994) does not mention a formal corporate governance structure. Also, Bangladesh is perhaps one of the few countries in the world without an ‘official’ code of corporate governance, although the Securities and Exchange Commission of Bangladesh (SEC) issued a brief corporate governance order only applicable for the publicly listed firms in Bangladesh (SEC 2006). Therefore, a vast majority of the Bangladeshi companies do not have any mandatory or ‘best practice’ guidance as far as corporate governance is concerned. Also, the adoption of SEC (2006) is not mandatory; rather, it is applicable on a ‘comply or explain’ basis.Footnote 3 The SEC order is in line with the western-styled shareholder model of corporate governance, and contains provisions such as a single-tiered board, presence of at least 10 % independent members, separation of CEO and the chairman, presence of an audit committee, etc. Very recently, the SEC (2012) has invited comments on proposals for reforming the SEC (2006) order. Major Proposals include increasing the number of independent directors in the board, specifying qualifications for independent directors, and enhancing the role of the audit committee.

Consistent with the findings from studies based in the context of emerging economies [for example, Lodhia (2003) in Fiji, Kuasirikun (2005) in Thailand], previous studies investigating CSR disclosure in Bangladesh have found such disclosure levels to be significantly low (for example, Belal 2000) and of rather descriptive nature, mostly reporting positive news (Imam 2000). Belal (2001) reported that a significant portion of the CSR-related disclosure involved information regarding employee-related information. The paper also questioned the credibility of such information, as the disclosures were mostly descriptive in nature. Azim et al. (2009), in a recent study of CSR disclosures by listed public companies in Bangladesh in 2007, revealed that only around 16 % companies made such voluntary disclosures. Consistent with Imam (2000) and Belal (2001), the paper also reported that the disclosures were mostly descriptive in nature, and no attempts for independent verification were made. The study also reported that the highest level of CSR disclosures prevailed in the banking sector.

Another stream of studies attempted to identify the perceptions and motivations for CSR disclosure in Bangladesh. Belal and Owen (2007) conducted an interview-based study to capture the views of corporate managers regarding the current state and future prospects of CSR reporting in Bangladesh. Interviews with 21 managers working in public and private sector entities and multinational companies operating in Bangladesh revealed the desire to satisfy key stakeholder groups to be the principal motivation for CSR reporting in Bangladesh. Managers working both in private and public entities identified government as a major stakeholder as it had the power to regulate the companies. Instructions from parent companies and concerns raised by buyer groups were also recognised as significant sources of pressure, especially for multinational companies and export-oriented industries. Other stakeholders mentioned by the respondents included media and influential lobby groups, and stakeholder groups such as the community, environment and wider society were either ignored or virtually unheard of. The role of ‘powerful’ stakeholder groups was also mentioned by Islam and Deegan (2008) as the principal driving force for CSR disclosures in Bangladesh. This is consistent with findings in other developing countries [for example, Rahaman et al. (2004) identify constant pressure from World Bank as a major influence for environmental reporting in Ghana].

In a recent paper, Belal and Cooper (2011) reported that Bangladeshi companies largely stayed away from more compelling CSR activities such as child labour, equal opportunities and poverty alleviation. Interviews with senior corporate managers identified absence of legal requirements, lack of awareness, shortage of resources and fear of bad publicity as the main reasons for such reluctance. Belal and Roberts (2010) expressed concerns that the motivation and practice for CSR in Bangladesh stems from international pressure, and the tendency to adopt stricter CSR regulations may lead to corruption or other unintended consequences, given the social and political environment prevailing in Bangladesh.

The Bangladesh Companies Act (GoB 1994) does not require mandatory disclosure of CSR activities for public limited companies. However, in 2008, the government issued a statutory regulatory order (SRO) that allowed companies to claim 10 % tax rebate on the actual amount spent on CSR activities (GoB 2008). The SRO identified specific economic, environmental and social development activities that would qualify for such exemptions. Recently, the GoB (2008) was replaced by a new SRO issue by the Bangladesh government (GoB 2011). The new order set a maximum threshold of Tk 80 million or 20 % of total income of the companies for CSR-related activities that qualified for the 10 % tax rebate.

Theoretical Framework

A number of previous studies (for example, Mukherjee-Reed 2002; Reed 2002; Siddiqui 2010) have suggested that adoption of western-styled models in emerging economies were promoted by attempts by governments and policy makers to gain legitimacy with external stakeholders such as international aid agencies and foreign governments. The legitimacy theory is also frequently used in the CSR literature to explain the motivations for CSR disclosures. The theory is based on the notion of a ‘social contract’, which limits the activities of an organisation within the boundaries set by the society (Gray et al. 1996). In essence, the organisation will gain support from the stakeholders and continue in existence in so far as its activities give benefits, or at least are not harmful to society. According to this theory, organisations continually seek to ensure that they are perceived as operating within the bounds and norms of their respective societies, that is, they attempt to ensure that their activities are perceived by outside parties as being ‘legitimate’. Perrow (1970) defines legitimacy as a generalised perception or assumption that the actions of an entity are desirable, proper or appropriate within some socially constructed system of norms, value, beliefs and definitions. Although firms have discretion to operate within institutional constraints, failure to conform to critical, institutionalised norms of acceptability can threaten the firm’s legitimacy and ultimately its survival (DiMaggio and Powell 1983; Oliver 1991; Scott 1987). An organisation therefore, through its top management, seeks congruency between organisational actions and the values of its general and relevant publics or its stakeholders (Dowling and Pfeffer 1975; Lindblom 1994). Sethi (1979) argues that if an actual or potential disparity exists between organisational and social values, then organisational legitimacy will be jeopardised giving rise to a legitimacy gap. A widening gap will cause an organisation to lose its legitimacy. Under such circumstance an organisation can adopt a number of public disclosure strategies for its survival (Lindblom 1994; Dowling and Pfeffer 1975).

In social and environmental accounting literature, many researchers concur that CSR disclosures can be employed by an organisation to mitigate legitimacy threat and reduce the legitimacy gap (see for example, Chen et al. 2008; Deegan et al. 2000, 2002). Legitimacy theory therefore implies that it is the top management of an organisation which is responsible to recognise legitimacy gap and carry out necessary social practices and disclose that accordingly to stakeholders to ensure accountability. Thus corporate governance, in particular the internal governance structure (such as ownership and board composition) is likely to play a vital role in reducing legitimacy gap through extended CSR disclosures.

Literature Review and Hypotheses Development

Managerial Ownership

The effect of managerial ownership on voluntary disclosures has been of interest to the accounting researchers for a long time. Largely, prior literature finds managerial ownership to be negatively related to levels of voluntary disclosure (for example, Eng and Mak 2003; Chau and Gray 2010). As mentioned before, owner-managed companies are very common in Bangladesh, and in most of the cases the board of directors comprise primarily of family members (Farooque et al. 2007). Such concentrated managerial ownership enables managers to dominate the company and decide upon the strategies and policies about organisational social behaviour. The dominance of family members in company management leads to the development of a tendency for important decisions to be first made in family meetings, and then regularised in formal board meetings, making such meetings largely symbolic (Ahmed and Siddiqui 2011). For this type of companies, public accountability may be less of an issue because outsiders’ interests may be relatively small. From the agency theory point of view, this is referred to as the ‘type II’ agency problem [see Villalonga and Amit (2006) and Kuo and Hung (2012) for a review of agency problems in family firms]. Additionally, because the level of public interest in closely held companies can be expected to be relatively low, this type of companies may be less active in social activities. In other words, managers of closely held companies may not invest heavily in socially responsible activities because the costs of investing in these activities may far outweigh its potential benefits. Hence less amount of CSR information can be expected in closely held or owner-managed companies. The findings of limited prior evidence also document a negative relationship between managerial ownership and the extent of CSR disclosures (Oh et al. 2011 ; Ghazali 2007). We therefore, propose the following hypothesis:

H1

Ceteris paribus, there is a negative association between managerial ownership and the level of CSR disclosures.

On the other hand, a number of recent studies (for example, Belal and Owen 2007; Islam and Deegan 2008) have identified important stakeholders, such as pressures from international buyers, as a critical factor for making CSR disclosures. This means that despite the presence of high managerial ownership, managers in export-oriented industries may have sufficient incentives to provide more CSR disclosures in order to allay any potential concerns of their foreign buyers. To test this claim, we propose the following sub-hypothesis in alternative form:

H1a

Ceteris paribus, there is a negative association between managerial ownership and the level of CSR disclosures in export-oriented industries.

Public Ownership

Prior studies also suggest that ownership dispersion across many investors contributes to increased pressure for voluntary disclosures (Chau and Gray 2002; Cullen and Christopher 2002; Ullmann 1985). The large number of stakeholders in a diffused company means that the benefits of disclosures are likely to outweigh the associated costs for the publicly owned firms. When a company is publicly held the issue of public accountability may become very important as well. Publicly owned firms are therefore expected to have more pressures to disclose additional information due to visibility and accountability issues resulting from large number of stakeholders. This may necessitate additional involvement in social or community activities and hence disclosure of these activities. Thus, it may be expected that public ownership concentration is positively associated with the extent of social activities. However, a number of prior studies have reported (for example, Siddiqui 2010; Uddin and Choudhury 2008) that shareholders in Bangladeshi companies are largely inactive. Sobhan and Werner (2003) mentioned that very few people attending the annual general meetings actually understood the financial statements being presented, and the main incentive for attendance was ‘the meal served in an attractive hotel’ (Sobhan and Werner 2003, p. 15). Therefore, it could be argued that wider public ownership in the context of Bangladesh may not result in greater CSR disclosures. Therefore:

H2

Ceteris paribus, there is a positive association between public ownership and the level of CSR disclosures.

Foreign Ownership

The demands for disclosures are generally higher when foreigners, due to the separation between management and owners geographically, hold a high proportion of shares (Schipper 1981; Bradbury 1991). Additionally, foreign investors are likely to have different values and knowledge because of their foreign market exposure. Thus, a company with foreign ownership is expected to disclose more information including social and environmental information to help them in decision making. Haniffa and Cooke (2005) find a positive significant relationship between foreign ownership and CSR disclosures in Malaysia indicating that Malaysian companies use CSR disclosure as a proactive legitimating strategy to obtain continued inflows of capital and to please ethical investors. It is, therefore possible that this group of investors can influence corporate disclosure practices including CSR disclosures of listed companies in Bangladesh. However, in Bangladesh, although foreign ownership is becoming increasingly common because of the growth in multinational ventures, the percentage of shares held by foreign investors are still very limited.Footnote 4 Therefore, it is also possible that foreign investors, due to the limited nature of their scale of investments in Bangladesh, may not be able to influence CSR disclosure practices. To test this, we therefore, propose the following hypothesis:

H3

Ceteris paribus, there is a positive association between foreign ownership and the level of CSR disclosures.

Board Independence

The presence of independent directors in the board is considered to be a major corporate governance mechanism. It is generally believed that independent outside directors will strengthen the board by monitoring the activities of the management, and ensure that interests of the investors are protected (Petra 2005). A number of studies (for example, Harjoto and Jo 2011) report CSR disclosures to be significantly correlated with board independence, measured in terms of the percentage of independent directors in the board. Consistent with this, it can, therefore, be argued that the presence of independent directors in the board will enhance CSR disclosures.

In the context of Bangladesh, however, the term ‘independent directors’ may have an entirely different meaning. Consistent with most western-styled corporate governance guidelines, the current SEC order (SEC 2006) requires at least 10 % of the members of the board of directors to be independent.Footnote 5,Footnote 6 However, Sobhan and Werner (2003) point out that a director who fits into the definition of being ‘independent’ in Bangladesh is often a former bureaucrat, appointed with the purpose of obtaining licenses or ‘as a payback for previous favours’. The paper goes on to make the following observation:

Very few independent directors are appointed for their expertise and the priority in appointing directors is usually their personal connections to company management or having connections that can be used for the company in the future. When boards need an independent opinion they rely on employing outside consultants or advisors. Therefore, in the context of Bangladesh, independent directors do not serve as an advocate for minority shareholders or as a source of new and different ideas. (Sobhan and Werner 2003, p. 35)

This is consistent with Uddin and Choudhury (2008), who find that in most Bangladeshi companies, independent directors were appointed for ‘name only’, and personal connections, rather than skill and expertise, were the major criteria for such selection. A recent study by the World Bank (2009) also identified the need for greater board independence and professionalism in the public limited companies in Bangladesh. Under such circumstances, the influence of independent directors in the board is doubtful. To test whether board independence has an influence on CSR disclosure, we develop the following hypothesis:

H4

Ceteris paribus, there is a positive association between proportion of independent directors and the level of CSR disclosures.

Role Duality

The combination of CEO and chairman positions results in leadership and governance issues. Li et al. (2008) argue that separation of the roles of chairman and CEO may enhance monitoring quality in critical decisions about stakeholder responsiveness. Haniffa and Cooke (2005) offer two views in this aspect. The first view supports the separation of the two roles to provide checks and balances for the performance of management, while the second view argues that the separation is not crucial since many companies are well run even with the roles combined and have a strong and capable board for monitoring. Nevertheless, companies with the CEO duality offer greater power to a person, which may enable him to make decisions that, do not take into account of greater interests of stakeholders. This may result in negligence of additional involvement in social or community activities and hence disclosure of these activities.

In Bangladesh, the SEC guidelines (2006) require the positions of the chairman of the board and the CEO to be filled by different individuals. However, unlike some corporate governance codes adopted in other countries [for example, the combined code in the UK (FRC 2011)], the chair of the board is not required to meet the independence criteria. Given the overwhelming presence of family dominance in the Bangladeshi corporate sector, it is rather unsurprising that such positions are very often filled by two members of the same family. Sobhan and Werner (2003) report that an overwhelming majority (73 %) of the non-bank listed companies interviewed in the survey, the board is heavily dominated by sponsor shareholders who generally belong to a single family, and ‘the father as the Chairman and the son as the managing director (CEO) is the norm’ (Sobhan and Werner 2003, p. 34). This is consistent with Uddin and Choudhury (2008), who report with the exception of multinational companies, most of the listed companies in Bangladesh are dominated by family members, and head of the family becomes the chairman and other family members occupy important posts such as the CEO or the managing director. Given this context, it would be interesting whether the provision for a separation of CEO and chair would have an effect on CSR disclosures. To test this, we propose the following hypothesis:

H5

Ceteris paribus, there is a negative association between CEO/chair role duality and the level of CSR disclosures.

Audit Committees

The corporate governance effects of audit committees have been well documented (for example, Turley and Zaman 2004, 2007). Audit committees comment on and approve choices of accounting policies, and are therefore, expected to influence a company’s approach towards financial reporting and disclosure. Pomeroy and Thornton (2008) conduct a meta-analysis of audit committee’s effects on financial reporting quality. The study reports a positive relationship between audit committee existence and financial reporting quality in emerging economies [for example, Bradbury et al. (2006), in Singapore and Malaysia; Rahman and Ali (2006) in Malaysia].

In the context of Bangladesh, SEC (2006) requires all listed companies to have audit committees comprising of at least three members of the board of which at least one must be an independent director [SEC (2012) proposes at least two members to be independent directors]. The order also requires the chair of the audit committee to have a ‘professional qualification or knowledge, understanding and experience in accounting and finance’ (SEC 2006, p. 4). Although the influence of audit committees on financial reporting quality has not been investigated, Ahmed and Siddiqui (2011), in a recent study report that corporate culture created due to family dominance tends to undermine the effects of the audit committee. Interview evidence seems to suggest that there is a corporate culture to undermine the role of the audit committee within the organisation. The study reports the opinions of one member of the board of directors of a commercial bank, who also holds majority of the shares:

The audit committee is like an ornament for my bank…I do not think they have any role in the functioning of the bank, nor they have anything to do with the auditors. They are there simply because it is a regulatory requirement…I have heard that in many companies, audit committee members are actually wives/family members of the directors who do not know much about the business. (Ahmed and Siddiqui 2011, p. 22)

The study also reports that family connections tend to override skill and competence as selection criterion for members of the audit committee. Hence, in the context of Bangladesh, audit committees may not have a significant influence on the corporate disclosure practices. Therefore we propose the following hypothesis:

H6

Ceteris paribus, there is a positive association between presence of audit committees and the level of CSR disclosures.

Research Design

Sample

The sample selection procedure is reported in Table 1. The sample consists of all 135 manufacturing companies listed with Dhaka Stock Exchange (DSE) in Bangladesh from 2005 to 2009.Footnote 7 We exclude 19 companies due to missing or incomplete information. The final sample comprises the remaining companies (116) with a total of 580 firm-years observations. The sample consists of various sectors such as: cement (7), ceramics (4), engineering (19), food (21), jute (3), paper and printing (2), miscellaneous (11), pharmaceuticals (21), tannery (5), paper and printing (2) and textile (23).The data for our analysis comes from multiple sources. We collect the financial and ownership data from the annual reports of the sample companies listed on the Dhaka stock exchange. Social responsibility information was hand-collected from the CSR disclosures, corporate governance disclosures, directors’ report, Chairman’s statement, and notes to the financial statement contained in annual reports.

Table 1 Sample description
Full size table

Model Specification

We use regression analysis to test the relationship between the corporate governance variables and the CSR disclosure. The assumptions underlying the regression model were tested for multicollinearity based on the correlation matrix as well as the variance inflation factor (VIF).Footnote 8 The regression equation is as follows:

$$ \begin{aligned} {\text{CSRDI}} = & \alpha + \beta_{1} {\text{MOWN}} + \beta_{2} {\text{PUB}} + \beta_{3} {\text{FOROWN}} + \beta_{4} {\text{BIND}} + \beta_{5} {\text{CEODU}} + \beta_{6} {\text{AUDCOM}} \\ & \, + \beta_{7} {\text{FSIZE}} + \beta_{8} {\text{FAGE}} + \beta_{9} {\text{LEV}} + \beta_{10} {\text{ROA}} + \beta_{11} {\text{INDUSTRY DUMMIES}} \\ & \, + \beta_{12} {\text{YEAR DUMMIES}} + \varepsilon \\ \end{aligned} $$

where CSRDI, corporate social responsibility disclosure score/index; MOWN, percentage of shares owned by the directors; PUB, percentage of shares owned by the public; FOROWN, percentage of shares owned by the foreign investors; BIND, proportionate indirect directors on the board; CEODU, dummy variable equals 1 if same person holds the positions of CEO and chairman in a firm; AUDCOM, dummy variable equals 1 if there is an audit committee in a firm; FISZE, natural logarithm of total assets; FAGE, natural log of the number of year since the firm’s inception; LEV, ratio of book value of total debt and total assets; ROA ratio of earnings before interest and taxes and total assets.

The corporate governance variables are managerial ownership (MOWN), public ownership (PUB), foreign ownership (FOROWN), proportion of independent directors on board (BIND), CEO duality (CEODU) and audit committee (AUDCOM). We also include control variables that have been found in prior research to be related to disclosure. The control variables included are firm size (FSIZE), firm age (FAGE), leverage (LEV) and return on assets (ROA).

MOWN is the proportion of ordinary shares held by directors. When managerial ownership is higher, directors are likely to be less concerned about public accountability, resulting in lower CSR disclosures (Ghazali 2007). PUB is the proportion of share own by the public. Higher level of public ownership makes the companies to be more accountable and there are pressures to disclose more information on CSR (Ullmann 1985). FOROWN is the proportion of foreign shareholders to total shareholders. Due to geographical separation between management and owners, foreign investors may demand higher disclosure on CSR (Bradbury 1991). BIND is the proportion of independent directors on board. Independent directors who are less aligned to management may be more inclined to encourage companies to engage in CSR disclosure (Chen and Jaggi 2000). CEODU refers to the same individual serving as both CEO and chairman of the board. CEO duality is more likely to be associated with lower levels of CSR disclosure since this may provide greater degree of CEO discretion and enable him less accountable for the greater interest of stakeholders (Haniffa and Cooke 2002; Gul and Leung 2004). The existence of audit committee (AUDCOM) is expected to improve financial reporting quality including greater level of CSR disclosures.

Larger firms are expected to disclose more information (Haniffa and Cooke 2005). Larger firms receive higher attention from the various groups in society and therefore would be under greater pressure to disclose their social activities to legitimise their businesses. Older firms provide more social responsibility disclosures (Roberts 1992). A more matured firm is concerned about its reputation and hence disclose more social responsibility information. Purushothaman et al. (2000) predict a negative relationship between LEV and CSR disclosures in that companies with high leverage may have closer relationships with their creditors and use other means to disclose social responsibility information. Profitable companies disclose more corporate social and environmental disclosures (Haniffa and Cooke 2005). Profitability allows management the freedom and flexibility to undertake and reveal more extensive social responsibility activities to general shareholders. Profitable companies demonstrate their contribution to society’s well being and legitimise their existence through disclosing social information. We also use year dummies and industry dummies for different sectors.Footnote 9

Dependent Variable—Corporate Social Responsibility Disclosure Index (CSRDI)

The CSR disclosure index (CSRDI) represents the dependent variable in this study. To assess the extent of CSR disclosure in annual reports, a checklist containing 20 items was constructed (see Appendix). We follow previous studies to construct this checklist. In particular we follow Haniffa and Cooke (2002, 2005) and Ghazali (2007) and develop a modified checklist including the items relevant to Bangladeshi companies. A dichotomous procedure is applied whereby a company is awarded 1 if an item included in the checklist is disclosed and 0 if it is not disclosed. We do not penalise a firm for non-disclosure if the item is not relevant to the firm. Such a judgement can be made after reading the entire annual report (Cooke 1992). Accordingly, the CSR disclosure index is derived by computing the ratio of actual scores awarded to the maximum score attainable (20) by that company (Ghazali 2007). The index is measured for each company as the ratio of the score obtained to the maximum possible score relevant for that company.Footnote 10

Consistent with prior disclosure index studies (Botosan 1997; Gul and Leung 2004), we use the Cronbach’s coefficient alpha (Cronbach 1951) to assess the internal consistency of our disclosure index. Internal consistency refers to the degree to which the items in a test measure the same construct. Botosan (1997) and Gul and Leung (2004) uses the coefficient alpha as a reliability statistics useful to assess the degree to which correlation among the information categories of the disclosure index is attenuated due to random error. The coefficient alpha for the five information categories in our disclosure index is 0.702. This statistics provides a good support that the set of items in the disclosure scoring index capture the same underlying construct.Footnote 11

Results

Table 2 provides the descriptive statistics for the variables used in the study. The average disclosure score is 0.223 (median = 0.20). The average managerial ownership (MOWN) is 27.40 %. The average level of public ownership (PUB) is 33.00 %. The average board independence (BIND) of our sample is 7.10 and 24.70 % of the CEOs in our sample firms are also the chairman of the board (CEODU).

Table 2 Descriptive statistics of dependent and independent variables
Full size table

Table 3 presents the correlation matrix among variables. Corporate social responsibility disclosure (CSRDI) score is negatively correlated with managerial ownership (MOWN) (ρ = −0.330). However, corporate social responsibility disclosure (CSRDI) score is positively correlated with foreign ownership (FOROWN) (ρ = 0.462) and board independence (BIND) (ρ = 0.269). CEO duality is positive but insignificantly correlated with CSRDI. The correlation between public ownership (PUB) and CSRD is also positive (ρ = 0.081). CSRDI is also correlated with the control variables such as firm size (FSIZE) (ρ = 0.558), firm age (FAGE) (ρ = 0.231), leverage (LEV) (ρ = −0.192) and return on assets (ROA) (ρ = 0.371).

Table 3 Correlation matrix
Full size table

Table 4 reports the mean values of the explanatory variables under analysis across the CSR disclosure scores for both firms with a score higher than the median and those with a score lower than the median. To test the statistical significance of the mean differences in the explanatory variables between both groups of firms, we perform a t test. It can be observed that firms with a CSR disclosure score higher than the median have higher foreign ownership (FOROWN), public ownership (PUB) and CEO duality (CEODU) and lower managerial ownership (MOWN) as compared to those firms with a CSR score lower than the median. Although firms disclosing more on CSR activities have higher board independence (BIND), this difference is not statistically different, at a 5 % level, between both groups of firms. Our analysis also shows that several control variables differ significantly between both groups of firms such as age (FAGE), leverage (LEV) and profitability (ROA).

Table 4 Differences in the value of the explanatory variables between firms with higher and lower CSRDI
Full size table

Table 5 reports the results of regressing the explanatory variables on the CSR disclosure score. In model 1 we examine the impact of managerial ownership on CSR disclosures. We find a negative significant coefficient (β = −0.114, p < 0.01) of managerial ownership (MOWN) variable. It implies that higher managerial ownership results in lower extent of CSR disclosure thus supporting H1. Our results also suggest that public interest as well as accountability could be less of an issue in owner-managed Bangladeshi companies. Owner managers may not be interested to invest heavily in CSR activities since costs associated with these activities may exceed the benefits. Thus, managers are relatively less concerned about social and environmental activities thereby reporting lower extent of CSR disclosures. This is consistent with a number of prior studies (for example, Ho and Wong 2001 ; Chau and Gray 2010) that find managerial ownership to be negatively related to levels of voluntary disclosure.

Table 5 Multiple regression results using CSRDI as the dependent variable
Full size table

In model 2 we explore the impact of managerial ownership on the extent of CSR disclosures in export-oriented industries. Bangladesh is currently the second largest exporters of textile products in the world. The textile industry is by far the most import industry in Bangladesh as it generates more than 80 % of the export earnings of the country. However, the industry and its large foreign buyers (such as Wal Mart, etc.) sometimes attract criticisms from the world press due to poor working conditions and levels of wages offered to the workers. Islam and Deegan (2008) report that multinational buyers of the textile industry now expect the manufacturers in Bangladesh to attend to various social and environmental issues, and failure to meet such expectations may lead to loss of supply contracts. Although, like most other industries in Bangladesh, the textile sector is also characterised by high levels of managerial ownership and family dominance,Footnote 12 pressures from international buyers may mitigate the influence of family ownership on CSR disclosure practices for this industry. Belal and Owen (2007) also identify pressures from international buyers in the garment industry in Bangladesh as an ‘extremely potent factor’ influencing CSR disclosure practices (Belal and Owen 2007, p. 485).

To test the effect of managerial ownership at the presence of international expectations to be socially and environmentally responsive, we create an interaction between managerial ownership and textile industry variables (EXPORT). We find a positive significant coefficient of the interaction (β = 0.247, p < 0.01) variable (MOWN*EXPORT) indicating a moderating effect of export-oriented industries on the negative relationship between managerial ownership and CSR disclosures. Thus our H1a is rejected. Our results imply that in spite of managerial ownership concentration management of export-oriented industries in Bangladesh report more CSR disclosures because of pressure from international buyers. This is consistent with and provides further empirical support to previous studies such as Belal and Owen (2007) and Islam and Deegan (2008), who identify pressures exerted by powerful stakeholder groups as a principal driver of CSR reporting in Bangladesh.

In the next model (model 3) we examine the impact of public ownership on CSR disclosures. We document a positive significant coefficient (β = 0.027, p < 0.05) of public ownership (PUB) variable. It suggests that higher percentage of public ownership results in higher level of CSR disclosures. This also supports H2. Our results also indicate that when a company is publicly held the issue of public accountability may become very important. These companies are expected to have more pressures to get involve in social or community activities. Therefore, publicly owned companies tend to get more involved in social activities to ensure organisational legitimacy and disclose more of these activities.

In model 4 we test the relationship between foreign ownership and the extent of CSR disclosures. We find a positive significant coefficient (β = 0.250, p < 0.01) of FOROWN (foreign ownership) which implies that higher percentage of foreign ownership results in higher extent of CSR disclosures, thus supporting H3. This implies that foreign investors are likely to have different values and knowledge related to contextual issues and are able increase strategic decisions in regards to public and social activities and their reporting. This is consistent with the findings of Haniffa and Cooke (2005) who suggest that companies with high foreign ownership report more CSR disclosures as a proactive legitimacy strategy to satisfy ethical foreign investors so that they get more foreign capital.

In model 5 we examine the relationship between board independence and the level of CSR disclosures. We find a positive significant coefficient (β = 0.372, p < 0.01) of board independence (BIND) variable. This is consistent with prior findings (for example, Petra 2005; Jo and Harjoto 2011). Our results thus imply that the greater the board independence, it is more likely that companies will emphasise on societal interests and organisational legitimacy (Haniffa and Cooke 2005) and disclose more CSR activities, supporting H4. Our results also support the notion that independent directors can put pressure on companies to engage in CSR to ensure congruence between organisational actions and societal values or organisational legitimacy. The results seem to imply that despite the common use of personal and family connections as the criteria for selecting independent directors, the presence of larger number of independent directors in the board helps diminish family influence on CSR disclosures.

We test the relationship between CEO duality and the extent of CSR disclosures in model 6. We document a positive but insignificant coefficient (β = 0.014, p > 0.10) of CEO duality (CEODU) variable. In other words it implies that CEO duality does not influence CSR disclosures for our sample companies. Thus it does not support H5. The finding seems to suggest that the custom of selecting the CEO and chair from the same family has turned the corporate governance mechanism of CEO/chair duality into a mere ritual.

In model 7 we investigate the impact of existence of audit committee on CSR reporting. We document a positive significant coefficient (β = 0.052, p > 0.01) of audit committee (AUDCOM) variable indicating that presence of an audit committee increases the extent of CSR disclosures. Thus, our H6 is supported which indicate that existence of an audit committee can ensure objectivity of financial reports through reporting a greater extent of CSR disclosures.

We regress CSR disclosure on all corporate governance variables in model 8 to test the impact of all the hypothesised variables in one model. Our results with respect to the coefficients of hypothesised variables are consistent with main findings reported in models 1–7. In regards to control variables, our overall findings suggest that larger firm size (FSIZE), older firms (LAGE) and better performance (ROA) are significantly related to greater extent of CSR disclosures. However, we find a negative significant impact of leverage on the level of CSR disclosures. The results of our analysis with respect to the control variables are consistent with the previous studies (Roberts 1992; Haniffa and Cooke 2005; Ghazali 2007).

A series of tests were conducted to test the model’s robustness. First, in model 6, we replace managerial ownership (MOWN) with percentage of ownership by family directors (FOWN) and board ownership (excluding family directors’ ownership). In the unreported results we find that family (FOWN) and board ownership (BOWN) (excluding family directors’ ownership) are significant and negatively related to CSR disclosures. Our results in regards to family ownership imply that family owners, who play a more dominant role in regards to corporate social strategies than other investors, are less concerned about public accountability and organisational legitimacy. Therefore, companies report less when there is high family ownership. The results remain unchanged in respect of other variables. Second, an OLS regression test was conducted by using the natural logarithm value of the CSR disclosure scores as the dependent variable. We rerun all the models (1–6) and document that our results do not differ qualitatively from those contained in Table 5. Third, an OLS regression test was also conducted by dropping all control variables from the model. Our results are consistent with the findings reported in Table 5. Fourth, we replace foreign ownership by foreign director dummy variable. Foreign director dummy variable is a dummy variable if there is any foreign director on the board. Thereafter, we rerun our model. We document a positive significant coefficient of the dummy variable. Once again it implies that because of their international exposure and knowledge foreign directors are more committed to protect the interest of the society and hence ensures more reporting of CSR disclosures. Finally, we partition our sample into two different sub-samples based on time periods—from 2005 to 2006 and from 2007 to 2009 and replicated the original analysis. The purpose of partitioning the sample is to test any impact of the ‘corporate governance notification 2006’ that took place during our study period. The results that we document for the sub-sample periods are qualitatively similar to the results in respect of the whole sample. Since the time period of our study could be influenced by increasing interest about CSR matters in global economy we attempt to test the presence of such increasing trends in our analysis. We run our model separately for each year and use a Chow test to evaluate the shifts of independent variables over time. Cross-period Chow tests of the significance of all the explanatory variables suggest no significant shift of coefficients over time.Footnote 13

Conclusions

The objective of this paper was to investigate the influence of corporate governance mechanisms on CSR disclosures in an Asian emerging economy, Bangladesh. The cultural diversity and political and legal structures in many developing countries (particularly, in the South-East Asian region) have significantly influenced the corporate governance practices in those countries. Consequently, most of these countries have corporate governance systems embedded on close, usually family relationships (Clarke 2007). Such family ownership and control has been enhanced through pyramidal structure and cross holdings (Claessens et al. 2000; Claessens and Fan 2002). Like many other countries family ownership is very common in Bangladesh and the Bangladeshi controlling families hold their shares independently in a particular company or group of companies. This ownership structure is, however, by no means similar to the pyramidal structure found in Western Europe and some other South-East Asian countries (Farooque et al. 2007).

Like some other developing economies (for example, India, Turkey, etc.), family firms are the most dominant form of business enterprises in Bangladesh. The strong family presence in the board of directors has led to the emergence of a culture where the values of corporate governance mechanisms are not always properly appreciated by the management. Prior research [for example, Mukherjee-Reed (2002) in India; Uddin and Choudhury (2008) in Bangladesh; Chen and Nowland (2010) in different Asian countries] has generally questioned the potency of corporate governance mechanisms at the presence of strong family control. In the case of Bangladesh, measures such as separation of chairman and CEO are likely to be affected by the fact that both these individuals are generally picked from the same family, and the independence and competence of the board of directors and the audit committee are likely to be hampered by the tendency to select independent members on the basis of personal connections rather than skill and competence. Prior studies on CSR disclosures in Bangladesh suggest that pressures from powerful stakeholders, rather than efficiency incentives, are the drivers for such disclosures. Under such environmental settings, the influence of corporate governance mechanisms on CSR disclosures is likely to be different from those in developed economy settings.

We document that managerial ownership is negatively related to the extent of CSR disclosures. This implies that high ownership concentration by the managers influence companies to get less involved in social activities because of their dominance than other investors thereby reporting less on CSR disclosures. However, we find that despite high level of managerial ownership, sample companies in export-oriented industries in Bangladesh make significantly more CSR disclosures. This provides further empirical support to prior interview-based findings by Belal and Owen (2007) and Islam and Deegan (2008) that the most potent factor for CSR disclosures in export-oriented industries in Bangladesh is the pressure exerted by important stakeholder groups. This also extends the work of Roberts (1992) and Dentchev (2004) by providing evidence that in the context of emerging economies, CSR disclosures are used by managers as a strategic tool to attain legitimacy. We document that public ownership and foreign ownership (including presence of foreign nationals in the board) have positive impacts on the extent of CSR disclosures. This means that despite lack of shareholder activism well dispersed firms in Bangladesh still provide significantly more CSR disclosures. The positive effect of foreign ownership and presence of foreign nationals in the board on levels of CSR disclosures reinforce our findings that managers in public limited companies strive to attain legitimacy by making higher amount of CSR disclosures.

We find corporate governance mechanisms such as board independence, and presence of an audit committee to have significant positive influence on the levels of CSR disclosures. However, we do not find any significant association between CEO/chairman role duality and levels of CSR disclosures. It may be noted that in the context of Bangladesh, CEO/chair role duality may not mean much as in most cases, these two roles tend to be occupied by individuals from the same family. It appears that despite the traditional settings, corporate governance mechanisms involving presence of outsiders have significant impact on the extent of disclosures made by Bangladeshi companies, possibly due to the legitimisation effects of such mechanisms. This is consistent with a number of prior studies (such as Chau and Gray 2010; Filatotchev et al. 2005) that find the presence of rational corporate governance mechanisms (such as board independence and independent chair) to have some mitigating effect on the influence of family ownership on voluntary disclosures. Perhaps, this suggests that corporate governance models require some further revisions requiring the chair of the board to be independent, or not allowing the chair and the CEO from the same family. The findings may have significant policy implications in the current context of Bangladesh, as the regulators are attempting to carry out further reforms in the corporate governance arena.

An alternative explanation could be that the presence of ethical management in some organisations influences its CSR disclosure practices. Prior studies (such as Fukukawa et al. 2007; Haniffa and Hudaib 2007) suggest a positive relationship between the ethical identity of a company and its corporate disclosures. In the context of Bangladesh, Belal and Roberts (2010) report that some managers and owners of some companies are more committed to social and environmental performance, and consequently, encourage more CSR disclosures. It could be argued that an ethical management would promote corporate governance mechanisms such as greater board independence and audit committee and at the same time, be more socially responsible.

The overall findings of our study provide empirical evidence which suggests that corporate governance attributes, albeit perceived to be less effective, are important determinants of extent of CSR disclosures in a developing country like Bangladesh. Thus corporate governance, in particular the internal governance structure, is likely to play a vital role to reduce legitimacy gap through environmental and social disclosures. Our study contributes to the literature by extending the findings of previous research which mainly focus on extent, content and motivational factors of CSR disclosures. Empirically, we provide evidence of managerial motivations for CSR disclosures in emerging economies. The findings of our study suggest that different types of owners may have divergent and even competing views on CSR practices. For example, managerial owners have a negative perspective toward CSR whereas public and foreign owners have positive perspectives. Such competing preferences may result in an inefficient CSR practices. Therefore, managers need to develop a sophisticated strategic decision making process to deal with competing preferences with respect to CSR practices. Our findings can also help the regulators and policy makers to adopt an appropriate balance of legislation, regulatory reform and their enforcement to make improvements in the corporate governance practices. Therefore, it should be of interest to regulators and policy makers in countries whose corporate ownership and regulatory structures are similar. While our results are probably dependent on Bangladesh’s institutional environment, learning the extent to which the results do generalise will help us better understand how institutional features matter for corporate governance and the extent of CSR disclosure. Thus, further studies in different jurisdictions on the issues we raise in this study are warranted.

The results of our study are subject to several limitations. Our study focuses on only disclosures in corporate annual reports although it is known that management may use other mass communication mechanisms. Therefore, future research may consider disclosures in other media such as newspapers, the internet, etc. Additionally, involvement in socially responsible activities may not necessarily translate into disclosure of those activities. The CSR disclosures index developed in this study may not have been fully or properly captured the CSR practices. Hence it should not be concluded that companies which did not disclose CSR information were not engaged in any social activities. The relationship between ethics, corporate governance, and CSR disclosure also needs to be explored. Finally, this study adopts the perspective of legitimacy theory, against the body of literature on corporate governance and disclosure practices that prefers agency theory. Further research may combine these two theories to develop a more comprehensive explanation about corporate governance attributes affecting the level of CSR disclosures.